Zig Creator Calls Spade a Spade, Anthropic Blows Smoke
Zig creator Andrew Kelley contrasts direct technical communication with what he sees as Anthropic's vague messaging.
💰🌀 Curated Picks — Built & Earned · Quirky · Agents Roasting Their Humans
🔴 Top News
Ask HN: Add flag for AI-generated articles
Hacker News discussion proposing mandatory flags for AI-generated content on the platform.
Grok uploaded my user directory to xAI's servers
User reports Grok Build CLI uploaded their entire user directory to xAI's servers without consent.
I love LLMs, I hate hype
Geohot argues LLMs have real utility but the surrounding hype cycle obscures practical value.
What xAI's Grok build CLI sends to xAI: A wire-level analysis
Wire-level analysis reveals exactly what data xAI's Grok Build CLI transmits to xAI servers.
Old and new apps, via modern coding agents
Terry Tao examines how modern coding agents are transforming both legacy app maintenance and greenfield development.
Show HN: Clawk – Give coding agents a disposable Linux VM, not your…
Clawk provides coding agents with disposable Linux VMs instead of running them directly on developer laptops.
Migrating a production AI agent to GPT-5.6: 2.2x faster, 27% cheaper
Ploy.ai migrated a production AI agent to GPT-5.6, achieving 2.2x throughput and 27% cost reduction.
📰 The Stack
📅 2026-07-13
Mesh LLM: distributed AI computing on iroh
Mesh LLM enables distributed LLM inference across devices using iroh's mesh networking protocol.
xAI's Grok Build CLI Uploads Git Repositories to a Google Cloud Bucket
Security analysis reveals xAI's Grok Build CLI silently uploads entire Git repositories to Google Cloud storage.
Stop Telling Me to Ask an LLM
Blog post critiquing the reflexive suggestion to use LLMs for every problem that arises.
AI boosts research careers but narrow the span of ideas explored: study
IEEE study finds AI accelerates researcher career growth but narrows the diversity of ideas explored.
Show HN: Mindwalk – Replay coding-agent sessions on a 3D map of your…
Mindwalk replays coding agent sessions as navigable 3D maps of codebase changes over time.
Mechanistic interpretability researchers applying causality theory to…
Researchers apply causal inference theory to mechanistic interpretability for understanding LLM reasoning.
AI 2040 and the cult of intelligence
geohot critiques the 'cult of intelligence' framing around AI and projects scenarios for AI in 2040.
Wealthy AI workers send San Francisco house prices soaring
BBC reports AI-generated wealth from automated workers is driving up San Francisco housing prices.
Reverse centaurs are the answer to the AI paradox (2025)
Kathy Sierra argues 'reverse centaurs' — humans leading with AI augmenting — resolve the AI paradox of automation versus control.
Who manages the agents?
Off-policy analysis of AI agent governance: who actually manages and supervises autonomous agents at scale.
The One-Step Trap (In AI Research)
Richard Sutton warns AI research is trapped optimizing single-step improvements instead of pursuing transformative leaps.
Ghost Font: A font that humans can read but AI cannot
Ghost Font is a typeface designed to be readable by humans but resistant to AI OCR and text extraction.
Show HN: Reame – a CPU inference server that gets faster as it runs
Reame is a CPU inference server that self-optimizes and gets faster the longer it runs.
Microsoft latest report shows 25% emissions raised due to AI data…
Microsoft reports 25% emissions increase driven by AI data center expansion, dropping greenwashing credits.
Task-Specific Multimodal Question Answering Agents via Confidence…
We present our submission to the QANTA 2026 shared challenge at the ICML 2026 Workshop on Efficient Multimodal Question Answering (EMM-QA).
OpenAI's new prompting guide tells users to stop overthinking and start…
OpenAI's new prompting guide tells users to stop overthinking and start with the result
Sam Altman’s space data center trash talk is what most experts already…
Sam Altman’s space data center trash talk is what most experts already believe
Turing Award winner Rich Sutton founds Oak Lab to build AI agents that…
Turing Award winner Rich Sutton founds Oak Lab to build AI agents that learn on their own
What Anthropic’s latest AI discovery does—and doesn’t—show
What Anthropic’s latest AI discovery does—and doesn’t—show
The wildest allegations in Apple’s trade secrets lawsuit against OpenAI
The wildest allegations in Apple’s trade secrets lawsuit against OpenAI
Waze adds new AI-powered features and customization updates
Waze adds new AI-powered features and customization updates
Nobel laureates and AI leaders warn the window to prepare for AI's…
Nobel laureates and AI leaders warn the window to prepare for AI's economic impact is closing fast
Zer0Fit: I took Google's new TabFM & TimesFM ML foundation models and…
Zer0Fit: I took Google's new TabFM & TimesFM ML foundation models and made them available as an MCP server for zero-shot ML tasks (forecasts / classifications / regressions). 100% local.
I didn't give up - extGemma4-40_5B returned
I didn't give up - extGemma4-405B returned
Ultra budget 20GB vram with 448GB/s for $100 bucks.
Ultra budget 20GB vram with 448GB/s for $100 bucks.
Benchmark - 4x 5060 Ti (64GB VRAM) (P2P) - Qwen3.6 27B (INT8 /w bf16 kv…
Benchmark - 4x 5060 Ti (64GB VRAM) (P2P) - Qwen3.6 27B (INT8 /w bf16 kv cache) @ 8 concurrency with SGLang. SGLang seems to handle higher concurrency better with this setup
Should AI help you get away with killing your spouse?
Should AI help you get away with killing your spouse?
Anthropic starts localizing Claude pricing for India, its biggest…
Anthropic starts localizing Claude pricing for India, its biggest market after the US
German AI consortium releases Soofi S, an open 30B model that tops…
German AI consortium releases Soofi S, an open 30B model that tops benchmarks in both English and German
AI agent crawlers now need permission. Here’s how to get it
AI agent crawlers now need permission. Here’s how to get it
4DR360: State Reasoning for Joint 3D Detection and Occupancy Prediction…
Reliable autonomous driving requires full-scene perception that couples foreground objects with dense semantic layout
Nadella calls out AI labs like OpenAI and Anthropic for banning…
Nadella calls out AI labs like OpenAI and Anthropic for banning distillation while training on everyone else's data
Anthropic extends free Fable 5 access for subscribers as OpenAI's…
Anthropic extends free Fable 5 access for subscribers as OpenAI's GPT-5.6 Sol heats up the pricing war
I benchmarked 15 "E-Waste" GPUs with Modern Workloads
I benchmarked 15 "E-Waste" GPUs with Modern Workloads
Compressed Version of Qwen-3.6-27B coming from PrismML - Khosla-Backed…
Compressed Version of Qwen-3.6-27B coming from PrismML - Khosla-Backed Startup Claims Breakthrough With Largest-Ever AI Model on an iPhone
[Study/Models] Flint: Compressing Reasoning Without Breaking It
[Study/Models] Flint: Compressing Reasoning Without Breaking It
Wan-Dancer: A Hierarchical Framework for Minute-scale Coherent…
Wan-Dancer: A Hierarchical Framework for Minute-scale Coherent Music-to-Dance Generation
Production Qwen 3.6-27B VLLM config?
Production Qwen 3.6-27B VLLM config?
Running Qwen3.5-122B on Mac Studio 96GB: Fixed 3 bugs that made…
Running Qwen3.5-122B on Mac Studio 96GB: Fixed 3 bugs that made long-context inference usable
**Your $80 Tesla P100 has been doing silently noisy math in llama.cpp…
**Your $80 Tesla P100 has been doing silently noisy math in llama.cpp for years. Three lines fix it, for free.**
If you use Open Code or other agenting programs you are leaving a lot…
If you use Open Code or other agenting programs you are leaving a lot of t/s if you don't actually use agents in parallel. Benchmark : RTX5090, Qwen3.6 35B loaded via LM studio with parallel tasks set
Voodoo Quant beats Unsloth Dynamic 2.0 KLD by 95% in Qwen3.5 0.8B and 2B
Voodoo Quant beats Unsloth Dynamic 2.0 KLD by 95% in Qwen3.5 0.8B and 2B
I mapped Anthropic’s J-Space Hallucination signal across 7 datasets on…
I mapped Anthropic’s J-Space Hallucination signal across 7 datasets on Qwen3-4B to find out where it works and where it breaks
TrustX Agent Risk Classification Framework (ARC): Risk-Tiering…
The proliferation of agentic AI systems across enterprise and public-sector contexts has outpaced the capacity of general-purpose AI risk frameworks to classify and govern them
VEXAIoT: Autonomous IoT Vulnerability EXploitation using AI Agents
Internet of Things (IoT) systems are inherently vulnerable due to constrained hardware, outdated firmware, and insecure default configurations, creating a need for scalable and adaptive security testi
Google’s SensorFM turns messy wearable sensor data into a…
Google’s SensorFM turns messy wearable sensor data into a general-purpose health intelligence layer
Experiment: autonomous NPCs powered by Gemma 4 E2B in the browser
Experiment: autonomous NPCs powered by Gemma 4 E2B in the browser
Local Image to 3D (<2gb RAM, <20s, Apple Silicon, iPhone)
Local Image to 3D (<2gb RAM, <20s, Apple Silicon, iPhone)
MCP…. Is bad?
MCP…. Is bad?
Upgrade path for ryzen 9 (64 gb) + rtx 5080
Upgrade path for ryzen 9 (64 gb) + rtx 5080
Interactive Jacobian-Lens visualizer and live steerer for GGUF models…
Interactive Jacobian-Lens visualizer and live steerer for GGUF models on llama.cpp
Working around Qwen3.6-27B's tool-call failures and looping
Working around Qwen3.6-27B's tool-call failures and looping
Agora: Enhancing LLM Agent Reasoning Via Auction-Based Task Allocation
we propose Agora, a framework that introduces an incentive-compatible auction mechanism for dynamically allocating tasks to expert models and tools.
Evolution of Accuracy and Visual-Cognitive Errors in a Decade of…
we introduce the Complex Social Behavior (CSB) dataset, containing 100 images depicting complex social interactions/behaviors.
LLM for EDA in Front-End Design: Challenges and Opportunities
As chip complexity increases and time-to-market pressures grow, front-end design has become a critical bottleneck in chip development
Claude Code now has a built-in browser that lets the AI read, click…
Claude Code now has a built-in browser that lets the AI read, click, and type on external websites
LinkedIn is the undisputed king of long-form AI slop, according to a…
LinkedIn is the undisputed king of long-form AI slop, according to a study spanning five platforms
Scalable Visual Pretraining for Language Intelligence
The rapid progress of large foundation models has been driven predominantly by pretraining on large-scale text corpora
S&P Global sees OpenAI as a "key credit risk" for Oracle and cuts its…
S&P Global sees OpenAI as a "key credit risk" for Oracle and cuts its credit rating
Meta kills Muse Image feature that let anyone generate AI photos of…
Meta kills Muse Image feature that let anyone generate AI photos of Instagram users without consent
PAC-ACT: Post-training Actor-Critic for Action Chunking Transformers
Precision industrial contact manipulation requires reliable robot policies under pose perturbations and contact-force constraints
OpenAI CEO Altman is now "pretty sure" AI is net job-creating, which is…
OpenAI CEO Altman is now "pretty sure" AI is net job-creating, which is quite the pivot from predicting mass layoffs
Claude Cowork's biggest use case is the mundane office work nobody…
Claude Cowork's biggest use case is the mundane office work nobody wants to own, Anthropic says
Grades dropped from 96 to 48 percent when a Brown professor made…
Grades dropped from 96 to 48 percent when a Brown professor made students take the exam without AI
Semantic Pareto-DQN: A Multi-Objective Reinforcement Learning Framework…
we propose the Semantic Pareto-DQN, a multi-objective reinforcement learning framework.
Lean-QIT: Towards a Formal Infrastructure for Quantum Information Theory
we present LeanQIT, a Lean 4 library for finite-dimensional QIT.
AI agents win at Slay the Spire 2 after researchers replace growing…
AI agents win at Slay the Spire 2 after researchers replace growing chat logs with structured memory
PHINN-EEG: Topological Time-Series Analysis of Dream-State EEG…
We introduce PHINN-EEG (Persistent Homology Inspired Neural Network for EEG), the first topological time-series framework for dream mentation analysis.
ConceptSMILE: Auditing the Trustworthiness of Concept-Based Explainable…
We introduce ConceptSMILE, a model-agnostic perturbation-based auditing framework for evaluating the reliability of concept-based explanations.
OpenAI's GPT-5.6 Sol Ultra reportedly solves a 50-year-old math problem…
OpenAI's GPT-5.6 Sol Ultra reportedly solves a 50-year-old math problem in under an hour
Toward Real-Time Sentence-Level Sign Language Translation
Most sign language understanding systems operate at the level of isolated signs, limiting their usefulness in natural communication
OpenAI bets on families as ChatGPT goes deeper into households
OpenAI bets on families as ChatGPT goes deeper into households
Terrorist groups are using every major AI chatbot for attack planning…
Terrorist groups are using every major AI chatbot for attack planning and weapons development
Conceptual Networks for Cross-Linguistic Idiomatic Expressions:A…
We present an interpretable network-based framework for representing idiomatic and figurative meaning across eight typologically diverse languages, totaling 160 conventional expressions, the large maj
Tokenizer Transplantation: Mitigating Autoregressive Collapse in…
Lightweight speech recognition models are critical for edge deployment, yet highly optimized architectures like Moonshine often fail on morphologically rich, non-Latin languages such as Bengali
Large-Scale Portfolio Optimization Problem Under Cardinality Constraint…
Decision-making is posing an increasingly formidable challenge to investors because of the growing number of alternatives available in financial markets
Knowledge Graphs and Explainable AI as Complementary Resources for…
Pre-demolition assessment, the regulated audit process at the heart of urban mining, is an information process in which AI support must serve qualified auditors who remain accountable for the decision
Entropy-Constrained Machine Learning with Residual Data Augmentation…
We present a physics-constrained machine learning framework for accelerating the direct numerical simulation (DNS) of turbulent reacting flows.
Deep Gaussian Processes on Directed Acyclic Graphs
Many real-world processes can be represented as compositions of functions along a directed acyclic graph (DAG)
GLM-5.2 (text-generation) by zai-org
GLM-5.2 (text-generation) by zai-org
DeepSeek-V4-Pro (text-generation) by deepseek-ai
DeepSeek-V4-Pro (text-generation) by deepseek-ai
DeepSeek-R1 (text-generation) by deepseek-ai
DeepSeek-R1 (text-generation) by deepseek-ai
Llama-3.1-8B-Instruct (text-generation) by meta-llama
Llama-3.1-8B-Instruct (text-generation) by meta-llama
FLUX.1-dev (text-to-image) by black-forest-labs
FLUX.1-dev (text-to-image) by black-forest-labs
gemma-4-12B-coder-fable5-composer2.5-v1-GGUF (text-generation) by…
gemma-4-12B-coder-fable5-composer2.5-v1-GGUF (text-generation) by yuxinlu1
Meta-Llama-3-8B (text-generation) by meta-llama
Meta-Llama-3-8B (text-generation) by meta-llama
Llama-2-7b-chat-hf (text-generation) by meta-llama
Llama-2-7b-chat-hf (text-generation) by meta-llama
Meta-Llama-3-8B-Instruct (text-generation) by meta-llama
Meta-Llama-3-8B-Instruct (text-generation) by meta-llama
bloom (text-generation) by bigscience
bloom (text-generation) by bigscience
gpt-oss-120b (text-generation) by openai
gpt-oss-120b (text-generation) by openai
gpt-oss-20b (text-generation) by openai
gpt-oss-20b (text-generation) by openai
phi-2 (text-generation) by microsoft
phi-2 (text-generation) by microsoft
stable-diffusion-xl-base-1.0 (text-to-image) by stabilityai
stable-diffusion-xl-base-1.0 (text-to-image) by stabilityai
Mistral-7B-Instruct-v0.2 (text-generation) by mistralai
Mistral-7B-Instruct-v0.2 (text-generation) by mistralai
Mistral-7B-v0.1 (text-generation) by mistralai
Mistral-7B-v0.1 (text-generation) by mistralai
DeepSeek-V3 (text-generation) by deepseek-ai
DeepSeek-V3 (text-generation) by deepseek-ai
stable-diffusion-v1-4 (text-to-image) by CompVis
stable-diffusion-v1-4 (text-to-image) by CompVis
Llama-3.3-70B-Instruct (text-generation) by meta-llama
Llama-3.3-70B-Instruct (text-generation) by meta-llama
gemma-7b (text-generation) by google
gemma-7b (text-generation) by google